Table 2 Summary of findings for studies testing the effect of the objective built environment on intervention-related physical activity outcomes (n = 20)
Author (year) | Study location | Sample design | Study design | Intervention details | Built environment measures included in analysis | Physical activity measures examined in relation to the built environment | Summary of key findings [Potential mechanism]a |
|---|---|---|---|---|---|---|---|
Carter et al. (2017) [ 32] | Leicester City/ Leicestershire County, UK (urban & rural) | At-risk of type 2 diabetes mellitus, 18–74 years Non-probability sample Analytical n = 706 | Two-arm RM RCT Arms: 1. Intervention 2. Standard care | Type: Individual (Walking away from diabetes) Exercise sessions: No Counselling: Yes Awareness/education: Yes Group-based activities: Yes Length: 3 sessions (at baseline, 12 & 24 months) | 800 m network buffer around home Walkability index (walkable area, road density, footpath density, junctions, cul-de-sacs, & connected node ratio). | Accelerometer Measured at baseline, 12, 24, & 36 months Domain(s): Non-specific PA type(s): MVPA Steps | “Arm x walkability index” interaction significantly associated with change in total steps & purposeful steps at 36-months. However, stratified analysis found no significant associations between walkability with change in total steps in the intervention (b = − 63.38, p > .05) or control (b = 19.68, p > .05) arms nor purposeful steps in the intervention (b = 42.97, p > .05) or control (b = 25.77, p > .05) arms. [Invariant ] |
Colom et al. (2020) [ 33] | Palma de Mallorca, Spain | Overweight/obese with metabolic syndrome, 55–75 years Non-probability sample Analytical n = 228 | Two-arm RM RCT Arms: 1.Full intervention 2.Partial intervention | Type: Educational with individual & group sessions (PREDIMED-Plus) Exercise sessions: No Counselling: No Awareness/education: Yes Group-based activities: Yes Length: 12 months (12 × 1 hr. one-on-one sessions, 12 telephone calls, 3 × 1 hr. group sessions) | 500 m/1000 m network sausage buffer around home Walkability index; residential density; intersection density; & land use mix | Accelerometer & self-reported Measured at baseline, 6 & 12 months Domain(s): Leisure/recreation Non-specific PA type(s): MVPA Walking | “Arm x walkability” interaction significantly associated with change in accelerometer-PA. Walkability associated with daily minutes of accelerometer-PA in intervention arm (b = 6.36, p < .05) but not control arm (b = 0.10, p > .05).[ Amplification ] Residential density associated with daily minutes of accelerometer-PA in intervention arm (b = 5.07, p < .05) but not control arm (b = 1.95, p > .05).[ Amplification ] Intersection density associated with daily minutes of accelerometer-PA in intervention arm (b = 6.86, p < .05) but not control arm (b = 1.38, p > .05).[ Amplification ] Land use mix not associated with accelerometer-PA in either intervention or control arm. [ Invariant ] |
Goyder et al. (2014) [ 34] | Sheffield, UK | Residents of deprived neighbourhoods (40–64 years), not achieving recommended levels of activity & wishing to have more support for PA Non-probability sample Analytical n: mini booster n = 47, full booster n = 52, control = 61 | Three-arm, parallel arm RM RCT Arms: 1. Full intervention 2. Partial intervention 3. Control | Type: PA consultations, using motivational interviewing (Sheffield Physical Activity Booster Trial) Exercise sessions: No Counselling: Yes Awareness/education: Yes Group-based activities: No Length: Brief (motivational DVD mailed to participants) & 2 follow up monthly phone calls | Postcode centroid Pedestrian access to municipal parks, swimming pools, & gyms using network distance analysis | Accelerometer Measured at baseline, 3 & 9 months Domain(s): Non-specific PA type(s): Total PA | No significant interactions between any BE variables & intervention arm in relation to energy expenditure at 3-months (9-month results not tested) [ Invariant ] |
Hays et al. (2016) [ 35] | Indianapolis, Indiana, USA | Inner city with risk of diabetes, ≥18 years Non-probability sample Analytical n = 216 | Two-arm RM RCT Arm: 1. Full intervention 2. Partial intervention | Type: Brief counselling plus a grouped-based diabetes prevention program (RAPID) Exercise sessions: No Counselling: Yes Awareness/education: Yes Group-based activities: Yes Length: 24 months (16 × 60–90 min sessions) | Walkability index (Walk Score®) & Normalized Differential Vegetation index (NDVI) at census tract-level | Accelerometer Measured at baseline, 6, 12, & 24-months Domain(s): Non-specific PA type(s): MVPA | No “BE x intervention interaction” tested. Separate estimates for intervention arm not presented (i.e., full & partial intervention pooled estimates). Higher NDVI was associated with increases in MVPA over time (b = − 115.03, p < .05). [ Amplification ] |
Kerr et al. (2010) [ 36] | San Diego County, USA | Overweight women (18–55 years) & men (25–55 years) Non-probability sample Analytical n: men = 89/ women = 83) | Two-arm pre-post RCT Arms: 1. Intervention 2. Standard care (some differences in interventions delivered to men vs. women) | Type: Computer-facilitated personalized action plan with brief counselling (PACE) – intervention delivery differed by sex. Exercise sessions: No Counselling: No Awareness/education: Yes Group-based activities: No Length: 12 months | 1600 m network buffer around home. Walkability index (residential density, intersection density, land use mix, retail floor area ratio). | Self-reported Measured at baseline & 12 months. Domain(s): Non-specific PA type(s): Walking | Men: “Time x arm x walkability” interaction significantly associated with change in daily walking duration. Intervention arm in low walkable neighbourhoods increased walking by 29-min/day [ Compensation ] versus intervention arm in high walkable neighbourhoods where walking decreased by 10-min/day [Suppression]. No change in walking for control arm regardless of neighbourhood walkability Women: No significant “time x arm x walkability” interaction. [ Invariant ] |
King et al. (2017) [ 37] | 4 sites: Texas, California, Pennsylvania, & North Carolina, USA | Sedentary, at major risk for mobility disability, 70–89 years. Non-probability sample Analytical n = 400 | Two-arm pre-post RCT Arms: 1. Intervention 2. Health education | Type: Centre-based (supervised) & home-based aerobic activities, strength training, flexibility, & balance exercises (LIFE-Pilot) Exercise sessions: Yes Counselling: Yes Awareness/education: Yes Group-based activities: Yes Length: 12 months (adoption phase: 2 months × 3 × 40-60 min session/wk.; transitional phase: 4 months 2 × 40-60 min session/wk.; maintenance phase: 6 months x optional 1–2 × 40-60 min session/wk) | 250 m buffer around home. Walkability index (residential density, intersection density, land use mix, retail to floor area ratio) | Self-reported Measured at baseline & 12-months Domain(s): Leisure/recreation Transport/errands PA type(s): Walking | No significant “arm x walkability” interaction associated with exercise/leisure walking. [ Invariant ] “Arm x walkability” interaction approached significance for weekly duration of walking for errands (p = .07). In low walkable neighbourhoods, walking for errands decreased by 15.4 min/week in intervention arm and increased by 8.7 min/week in health education arm (p < .01 for between arm difference). No between arm differences in walking for errands among high walkable neighbourhoods. However, results figures showed similar declines in walking for errands in intervention arm for both low and high walkable neighbourhoods. [Invariant] |
Lee et al. (2012) [ 38] | Houston & Austin, USA | African American, Hispanic or Latino women, 25–60 years Non-probability sample Analytical n = 309 | Two-arm pre-post RCT Arms: 1 PA Intervention 2. Nutrition intervention | Type: Group/team facilitated session & activity with personalized behaviour change plans (Health is Power) Exercise sessions: Yes Counselling: Yes Awareness/education: Yes Group-based activities: Yes Length: 6 months (6 group sessions) | 400 m radial buffer around homes. Pedestrian environmental data scan (audit of street segments): features & facilities related to walking & cycling (e.g., indicators of land use, connectivity, lighting, articulation, road attributes, safety attractiveness, sidewalks, & amenities) | Self-reported & accelerometer Measured at baseline & 6-months Domain(s): Non-specific PA type(s): MVPA (accelerometer) Total PA (self-report) Walking (self-report) | “Arm x BE” interaction significantly associated with post-intervention self-reported walking (number of traffic control devices & number of amenities) & total PA (number of crossing aids & number of amenities), but not with accelerometer-PA. Positive association between number of traffic control devices and post walking in PA group relative to a negative association in nutrition group (b = − 0.80, p < .05) [Amplification in PA arm] Positive association between number of crossing aids and post total PA in PA group relative to a negative association in nutrition group (b = − 0.46, p < .05) [Amplification in PA arm] Negative association between number of amenities and post walking in PA group relative to a positive association in nutrition group (b = 0.45, p < .05) [Suppression in PA arm] Negative association between number of amenities and post total PA in PA group relative to a positive association in nutrition group (b = 0.33, p < .05) [Suppression in PA arm] |
Lo et al. (2019) [ 39] | Montana & New York, USA (rural) | Sedentary overweight women, ≥40 years. Non-probability sample Analytical n = 151 | Two-arm pre-post clustered RCT Arms: 1. Full intervention 2. Partial intervention | Type: Multi-component, class-based exercise with motivational interviewing & awareness raising (Strong Hearts, Healthy Communities) Exercise sessions: Yes Counselling: Yes Awareness/education: Yes Group-based activities: Yes Length: 6 months (2 sessions/wk) | Walkability index (Walk Score®) | Accelerometer Measured at baseline & 6 months Domain(s): Non-specific PA type(s): MVPA | “Arm x Walk Score®” interaction not significantly associated with post-intervention daily minutes of MVPA or percent of MVPA. However, in neighbourhoods with a Walk Score® = 0, change in minutes (11.7 min/day) & percent (1.5%) of MVPA was significantly (p < .05) higher in intervention arm than change in minutes (−0.8 min/day) & percent (− 0.1%) of MVPA in control arm. No other between arm differences found for other Walk Score® levels. [Likely Compensation] |
McGowan et al. (2017) [ 40] | Edmonton, Canada | Prostate cancer survivors, ≥18 years Probability sample Analytical n = 165 | Three-arm RM RCT Arms: 1. Intervention (self-administered implementation intention; SAII), 2. Intervention (telephone-assisted implementation intention; TAII) 3. Brief information (PA factsheet: SPAR) | Type: Individual-based, no group sessions, information provided, no PA or exercise sessions (PROMOTE). Exercise sessions: No Counselling: Yes Awareness/education: Yes Group-based activities: No Length: Single session | 500 m/1000 m network buffers around homes. Walkability index (intersection density, residential density, & land use mix). Count of sport complexes (included facilities for indoor/outdoor activities) | Self-reported Measured at baseline, 1 & 3 months Domain(s): Leisure/recreation PA type(s): Total PA | “Time x arm x BE” interactions not significantly associated with minutes of self-reported weekly PA nor achievement of 150 minutes/week. [ Invariant ] |
Michael & Carlson (2009) [ 41] | Portland, USA | Mobility non-restrictive, cognitively sound adults not enrolled in formal exercise programs, ≥65 years. Probability sample Analytical n = 582 | Two-arm RM clustered RCT Arms: 1. Intervention 2. Education only | Type: Led walking groups (SHAPE) Exercise sessions: Yes Counselling: No Awareness/education: No Group-based activities: Yes Length: 6 months (3 sessions/wk) | 400 m/800 m radial buffer around home. Walkability index (e.g., sidewalk coverage, connectivity, public transportation access, distribution of parks/green space, & level of automobile traffic volume). | Self-reported Measured at baseline, 3, & 6-months Domain(s): Non-specific PA type(s): Walking | “Arm x walkability” interaction not significantly associated with weekly minutes of brisk walking measured at 6-months. [ Invariant ] |
Riley et al. (2013) [ 42] | Ottawa, Canada | ≥1 modifiable CHD behavioral risk factor, 20–80 years, with relative hospitalized for CHD in past 5-years Non-probability sample Analytical n = 230 | Two-arm pre-post RCT. Arms: 1. Intervention 2. Standard care | Type: Individual counselling sessions with a personalized behavior change plan (Family Heart Health) Exercise sessions: No Counselling: Yes Awareness/education: Yes Group-based activities: No Length: 12 weeks (1 in-person counselling session, followed by 12 weekly telephone calls) | Spatially defined neighbourhoods Walkability index (Walk Score®) & Walkability index (intersection density, residential density, retail floor area ratio & land use mix) | Self-reported Measured at baseline & 12-weeks Domain(s): Leisure/recreation PA type(s): MVPA | “Arm x Walk Score®” interaction not significantly associated with meeting the PA guidelines at 12 weeks (≥150 min/week of MVPA). [ Invariant ] “Arm x walkability index” interaction not significantly associated with meeting the PA guidelines at 12-weeks (≥150 min/week of MVPA). [ Invariant ] |
Robertson et al. (2012) [ 43] | Glasgow, Scotland, UK | Insufficiently active, 18–65 years Non-probability sample Analytical n = 45 | Two-arm RM RCT Arms: 1. Full intervention 2. Partial intervention | Type: Pedometer facilitated individualized walking program (Walking for Well-Being in the West) Exercise sessions: Yes Counselling: Yes Awareness/education: Yes Group-based activities: No Length: 3 months | 400 m radial buffer around home but also tested 1600 m. Street audits using the Scottish Walkability Assessment Tool (SWAT) & GIS measures (green space & recreation facilities, commercial & residential land use mix, dangerous & busy roads, pathway features other than safety, pathway safety features, roads & bus stops indoor fitness facilities & traffic calming features, traffic signals & pedestrian signage). Factors created. | Pedometer Measured at baseline, 3, 6, & 12-months Domain(s): Non-specific PA type(s): Steps | Separate estimates for intervention arm not presented (i.e., full & partial intervention pooled estimates) Dangerous & busy roads factor was negatively associated with step count at 3-months (b = −0.28, p < .05) and 6-months (b = − 0.31, p < .05). [Amplification] Commercial & residential land use factor was positively associated with step counts at 6-months (b = 0.40, p < .05) and 12-months (b = 0.30, p = .05) [ Amplification ] Traffic signals & pedestrian signage factor was negatively associated with step count at 6-months (b = − 0.30, p < .05). [Amplification] Indoor fitness facilities & traffic calming features factor was positively associated with step counts at 6-months (b = 0.27, p < .05). [ Amplification ] Green space & recreation facilities factor was negatively associated with step counts at 12-months (b = − 0.34, p < .05) [ Suppression ] Pathway features other than safety factor, Pathways safety features factor, & Roads & bus stop factor not associated with steps at any period. [ Invariant ] |
Zenk et al. (2009) [ 44] | Chicago, USA | Sedentary African American women, no signs/symptoms of CVD, 40–65 years Non-probability sample Analytical n = 252 | Two-arm post measure quasi-experiment Arms: 1. Full intervention 2. Partial intervention | Type: Walking program, including walking prescription, plus motivational workshops (The Women’s Walking Program) Exercise sessions: Yes Counselling: Yes Awareness/education: Yes Group-based activities: Yes Length: 12 months (24 week adoption phase including weekly motivation workshops & 24 week maintenance phase including weekly or bi-weekly telephone calls) | 1600 m radial buffer around home. Walkability index (land use mix, street connectivity, housing unit density, public transit stop density); Aesthetics; physical deterioration; industrial land use; availability of outdoor facilities & spaces; percentage of recreational open space area; indoor facilities; presence of public recreation center with indoor track or treadmill & shopping mall. | Self-reported & heart rate monitoring Measured post-intervention (i.e., accumulated PA) Domain(s): Non-specific PA type(s): Prescribed walks completed | No significant interactions between arm & any BE variables related to adherence to the program during the adoption phase. [ Invariant ] Separate estimates for intervention arm not presented (i.e., full & partial intervention pooled estimates). Presence of either a public recreation center with treadmill or indoor track or indoor shopping mall within 5 miles was associated with higher walking adherence relative to no facility present (b = 0.39, p < .05). [Amplification] Walkability not significantly associated with walking adherence (b = − 0.12, p > .05). [Invariant] |
Barnes et al. (2013) [ 45] | Perth, Australia | General population, 20–54 years Probability sample Analytical n: pre-interventio n = 466, post-intervention = 360) | One-arm pre-post quasi-experiment (repeated cross-sectional surveys) All participants exposed to intervention | Type: Mass media ( Find Thirty Every Day ) Exercise sessions: No Counselling: No Awareness/education: Yes Group-based activities: No Length: 12 months | 1600 m network buffer around home Recreational walkability index (dwelling density, street connectivity, & land use mix) | Self-reported Measured at pre & post intervention Domain(s): Transport/errands Non-specific PA type(s): Total PA Walking | “Time x walkability” interaction not significantly associated with any PA outcomes. However, adults from low walkable neighbourhoods were less likely to report any transport walking post vs. pre intervention (OR = 0.7, p < .05) while there was no significant change among those in high walkable neighbourhoods (OR = 0.9, p > .05). [Likely suppression] Adults from low walkable neighbourhoods were more likely to report sufficient PA post vs. pre intervention (OR = 1.4, p < .05) while there was no significant change among those in high walkable neighbourhoods (OR = 1.6, p > .05). [Likely invariant] |
Consoli et al. (2020) [ 46] | Calgary, Canada | Inactive, ≥18 years Non-probability sample Analytical n = 466 | One-arm pre-post quasi-experiment All participants received intervention | Type: Internet-facilitated pedometer intervention (UWALK) Exercise sessions: No Counselling: No Awareness/education: Yes Group-based activities: No Length: 3 months | Walkability index (Walk Score®) | Pedometer Measured post-intervention (i.e., accumulated PA) Domain(s): Non-specific PA type(s): Steps | Walk Score® was not associated with count of days steps recorded (IRR = 1.0, p > .05), achievement of 10,000 steps/day (IRR 1.0, p > .05), or daily steps during the intervention (b = 4.0, p > .05). [ Invariant ] |
Garmendia et al. (2013) [ 47] | Santiago, Chile | Residents of low socioeconomic county that had more than 400 persons (65–67.9 yrs) Non-Probability sample Analytical n = 996 | Original design included a multi-arm RM experiment but results for only one arm included in the analysis. All participants received a PA intervention | Type: Physical activity group exercise sessions Exercise sessions: Yes Counselling: No Awareness/education: No Group-based activities: Yes Length: 24 months (2 × 1 hr. sessions/wk) | Predefined neighbourhoods in Observatory of the Government of Chile Distance from home to the PA center (where activities were undertaken), and well-kept community green areas (m2/inhabitant) | Recorded Measured at 24 months Domain(s): Leisure/recreation PA type(s): Attendance at PA sessions (≥24 sessions) | Distance from home to the PA center not significantly associated with adherence to intervention (OR = 1.0, p > 0.2). [ Invariant ] Adherence was positively associated with neighbourhood area of well-kept green spaces (OR = 1.2, p < .01) [ Amplification ] |
Goyder et al. (2016) [ 48] | Sheffield, UK | Sedentary, from SES deprived neighbourhoods, 40–64 years Non-probability sample Analytical n = 941 | One-arm pre-post quasi-experiment All participants received intervention | Type: PA consultations, using motivational interviewing (Sheffield Physical Activity Booster Trial) Exercise sessions: No Counselling: Yes Awareness/education: Yes Group-based activities: No Intervention length: Brief (motivational DVD mailed to participants) & 2 follow up monthly phone calls | Shortest pedestrian network distance to: greenspace, gyms & pools from home. | Self-reported Measured at baseline & 3-months Domain(s): Non-specific PA type(s): Total PA | BE variables were not significantly (p > .05) associated with the likelihood of increasing PA due to intervention (gym: OR = 1.0; greenspace: 0.9; pool: OR = 0.9). [ Invariant ] |
Hino et al. (2019) [ 49] | Yokohoma City, Japan | General population, ≥40 years Probability sample Analytical n = 2023 | One-arm pre-post quasi-experiment All participants received intervention | Type: Pedometer facilitated walking program (Yokohama Walking Point Program) Exercise sessions: No Counselling: No Awareness/education: Yes Group-based activities: Yes Length: ~ 30 months | Distance to nearest railway station from center of each participant’s neighbourhood, & neighbourhood bus stop density. | Self-reported Measured post-intervention (i.e., accumulated PA) Domain(s): Non-specific PA type(s): Relative change in daily step count due to program | Bus stop density was positively associated with an increase in step counts due to the program (b = 0.04, p < .01). [ Amplification ] Distance to nearest railway station was not associated with an increase in step counts due to the program (b = 0.02, p < .05). [Invariant] |
Hino et al. (2021) [ 50] | Yokohoma City, Japan | General population, ≥40 years Probability sample Analytical n = 47,233 | One-group pre-post quasi-experiment All participants received intervention | Type: Pedometer facilitated walking program (Yokohama Walking Point Program) Exercise sessions: No Counselling: No Awareness/education: Yes Group-based activities: Yes Length: ~ 53 months | Distance to nearest railway station, & neighbourhood population density & intersection density. | Pedometer Measured post-intervention (i.e., accumulated PA) Domain(s): Non-specific PA type(s): Steps | Participants living furthest from a railway station (quartiles 1) undertook fewer daily steps than those living closer (quartile 2, 3 & 4) in spring (differences = − 289,- 464, & -400, all p < .0001), summer (differences = 286, 498, 515, all p < .0001), autumn (differences = − 277, − 436, & -386, all p < .0001), & winter (differences = − 268, − 426, & -363, all p < .0001). [ Amplification ] Participants living in neighbourhoods with lowest population density (quartiles 1) undertook fewer daily steps than those in higher density neighbourhoods (quartiles 2, 3 & 4) in spring (differences = − 126, − 232, & -354, all p < .05), summer (differences = − 173, − 321 & -433 all p < .001), autumn (differences = − 154, − 272, & -378, all p < .01), & winter (differences = − 145, − 255 & -345, all p < .01). [ Amplification ] There were no significant differences in daily steps based on neighbourhood intersection density quartile in any season. [Invariant] |
Jilcott et al. (2017) [ 51] | Lenoir County, North Carolina, USA (rural) | Patients with high blood pressure, ≥18 years Non-probability sample Analytical n = 249 | One-arm pre-post quasi-experiment All participants received intervention | Type: Individual or group counselling (Healthy Heart Lenoir) Exercise sessions: No Counselling: Yes Awareness/education: Yes Group-based activities: Yes Length: 4 months (1 session/month) | 1600 m network buffer around home. Walkability index (Walk Score®), density & distance to closest PA venues (parks, trails, & gyms) & food environment | Self-reported activity logs, & pedometer. Measured at baseline & 6-months Domain(s): Non-specific PA type(s): Total PA Steps | Increased distance (miles) to private gyms was associated with a larger increase in 6-month change in total PA (b = 31.7, p < .05) [Suppression] Increased density of private gyms was associated with a lower increase in 6-month change in daily steps (b = − 491.1, p < 05). [Suppression ] |